Stop Using Fragile Chatbots: Build Standard MCP Agents with TypeScript

说明与用途

开发者可以构建本地安全的 AI 助手，在不将数据发送到云端的情况下执行真实操作，节省人工工作时间。

问题

LLMs are trapped in chat boxes and require brittle Zapier workflows or weeks of custom function‑calling code to perform real actions like reading files or triggering APIs.

结果

Developers can now create local, secure AI assistants that perform real actions on client data without sending it to the cloud, saving hours of manual work and reducing reliance on fragile workflows.

工作流

• 1. Define agent directory structure with plugin.json, mcp.json, and SKILL.md.
• 2. Use the Agent Orchestrator to generate the directory and files from a natural language prompt.
• 3. Apply normalizeMcpConfigForPlatform to wrap npx/uvx commands for Windows compatibility.
• 4. Deploy the agent locally (e.g., via start.bat) so clients can double‑click to run the assistant.
• 5. Use the web UI to input a task, parse intent, select the filesystem MCP server, and estimate token cost before execution.
• 6. Execute the agent, which performs file operations, categorizes PDFs, or drafts emails without sending sensitive data to the cloud.

证据

To make an LLM actually do things—like read a local file, trigger an API, or move a PDF—you either rely on brittle Zapier workflows or spend weeks writing custom OpenAI function‑calling code.

来源

If you've tried running the official @modelcontextprotocol/server-filesystem on Windows, you know the pain. Windows doesn't recognize npx as an executable in stdio transport. It expects a .cmd shim.

来源

The UI hits the backend API, parses the intent, and dynamically selects the filesystem MCP server. It also calculates the estimated token cost (using DeepSeek or GPT-4o) before you even run it.

来源

With a standardized MCP agent, you can now hand a client a folder. They double-click start.bat, and they have a local, secure AI assistant that manages their Excel sheets, categorizes their invoices, or drafts email replies—without sending sensitive data to a cloud database.

来源